Beyond Code: Engineering Trustworthy Software Systems with AI at Scale was authored by Randal Burns (Johns Hopkins University), Sebastian Elbaum (University of Virginia), Gabrielle Allen (University of Wyoming), Nils Aschenbruck (Osnabrück University), Terry Benzel (University of Southern California), William Gropp (University of Illinois Urbana-Champaign), Rick Kazman (University of Hawaii), and Manish Parashar (University of Utah).

Read the Report Here

Avoiding Security Breaches in AI-Generated Software

Software built using AI is entering systems that people depend on. But AI now generates code faster than anyone can review it, and the report finds that the resulting systems carry unclear provenance, configuration errors, unvetted third-party components, and expanded attack surfaces. Faults missed in AI-generated code are already causing security breaches.

Organizations also cannot fully tell what is in the code they receive. Because these systems are black boxes with respect to their training data, users are unable to determine whether generated code contains copyrighted material, infringes patents, or reproduces proprietary work belonging to someone else. Major providers have responded by offering their customers legal indemnification. But interestingly, the market is developing ways to allocate this risk faster than the technical means to actually measure it.

Meanwhile, the checks that would normally catch these problems are weakening:

• Reviewer fatigue. People facing large volumes of machine-generated code miss subtle faults and tend to defer to the machine
• Misleading test results. Automated testing can produce results that look thorough while confirming the wrong behavior
• Concentration risk. When most organizations rely on the same dominant model, they inherit the same blind spots at once

Because of this, verifying code has to scale the way generating code has; and it needs research funding to do so. The report calls for automated checking that runs at machine speed, including formal verification methods that prove a system meets its specification rather than testing samples of its behavior. It also calls for tools that trace generated code back to its sources and for auditing of training data that establishes clear licensing, so organizations can assess what they are deploying before they deploy it. Against concentration risk, it proposes using different AI models to probe each other’s output for vulnerabilities.

Who Can Afford to Study These Systems

Evaluating whether AI systems are safe requires running them at the scale they actually operate. The report finds that the ability to do that now sits almost entirely with industry, which controls the computing power, the training data, and the engineering talent, along with the deployment environments where these systems run in production. Universities increasingly study commercial products after release rather than building the next generation of systems themselves. The report describes academic researchers:

• Evaluating industry outputs rather than developing systems of their own
• Auditing systems they cannot see inside for bias and other properties
• Reproducing findings at a fraction of the scale available to industry labs
• Leaving for industry altogether

University labs still lead on safety questions, including the social and ethical effects of AI, but the report finds academic relevance declining elsewhere. It identifies one fix as fast-acting: giving academic researchers access to frontier computing infrastructure, through a combination of industry funding and national compute initiatives.

A Workforce Built for Different Work

The report finds that the job of a software developer is changing, from writing code toward specifying what a system should do and supervising the AI tools that produce it. Industry has already started hiring on that basis, evaluating engineers on system design and their ability to direct and verify AI output. That shift outpaces the training pipeline. In higher education, the report calls for:

• Curricula centered on requirements, specification, and system design rather than isolated programming exercises
• New teaching approaches, since students will no longer build their understanding of systems through the act of programming
• Retraining beyond the university, reaching the current workforce whose implementation-focused roles are disappearing

Read the Full Report

The findings of this report are informed by discussions at the CCC Beyond Code: Engineering Trustworthy Software Systems with AI at Scale visioning workshop held February 25-26, 2026 in San Francisco, California, generously supported by the National Science Foundation (NSF) and the IEEE Computer Society. The workshop convened 41 experts across academia, industry, and government in artificial intelligence, software engineering, programming languages, cybersecurity, and systems engineering.

Read the Full Report

Tune in to the CCC LinkedIn Showcase Page for updates and more reports like this. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.

This material is based upon work supported by the U.S. National Science Foundation (NSF) under Award Nos. 2300842 and 2619366. These awards support the Computing Community Consortium (CCC), a programmatic committee of the Computing Research Association (CRA). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

The workshop will take place October 13-15, 2026, at the Chaminade Resort in Santa Cruz, CA. Over two and a half days, participants will take part in an intensive, laptop-free program focused on consensus-building, working together to define strategic directions and priorities for the AI research ecosystem based on community input.

Understanding the Stressors on AI Research

AI technologies are advancing quickly, and deployed AI systems have become major drivers of economic and social change. Yet the AI research ecosystem itself has also been evolving rapidly.

“In a world increasingly focused on adoption of AI with its myriad uses, potential misuses and concerns around AI security, there is a renewed need to consider all the stressors on the AI Research Ecosystem so that much needed and well thought out AI research can thrive,” says workshop co-chair Vandana Janeja, Professor of Information Systems at the University of Maryland Baltimore County. “I am hopeful that this workshop will lead to recommendations on how best to support this ecosystem into the next several years.”

The workshop’s specific focus areas emerged from months of community input. Through a Consider.it forum, researchers could submit stressors they saw affecting the AI research ecosystem and respond to those raised by others. In parallel, the AIRE Task Force convened virtual roundtables by sector, bringing together perspectives from government (including DoE, NSF, NIH, and NASA), industry (including Microsoft Research), professional societies (including AAAI), and academia (including Carnegie Mellon University and Brown University). The findings from both efforts led to four pillars, which participants will explore in depth at the workshop: 

• Publication and peer review: Conference reviewing systems, journal publishing systems, and pre-publication distribution
• Infrastructure and resources: The computing, data, and funding resources, including Federal funding programs, that make AI research possible
• Interdisciplinarity: How AI research connects with and draws on other fields
• Public-private partnerships: Research translation, translational ecosystems, and standard agreements for industry funding of academic research

“While AI offers amazing opportunities for the academic AI research community, it is also straining the academic research ecosystem, including publication and review structures, academic reward structures, research infrastructure, and partnerships,” says workshop co-chair Manish Parashar, Chief AI Officer at the University of Utah. “I look forward to the AIRE visioning workshop, where we will come together to understand these stressors, brainstorm mitigation strategies, and develop an action plan to restore a vibrant AI research ecosystem.”

Learn More and Stay Up-to-Date

The final sessions of the workshop will focus on synthesizing these discussions into recommendations, ultimately informing a CCC report expected in Spring 2027. 

CCC looks forward to welcoming participants to the workshop. To stay informed about updates and outcomes from the AIRE workshop, follow the CCC Blog and LinkedIn Spotlight Page. We also recommend checking out Janeja and Parashar’s recent Medium post, “Envisioning the Future AI Research Ecosystem” for more information about their initial findings from the field.

This material is based upon work supported by the U.S. National Science Foundation (NSF) under Award No. 2300842 and 2619366. These awards support the Computing Community Consortium (CCC), a programmatic committee of the Computing Research Association (CRA). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

SEMANTiCS is a globally recognised venue where researchers, industry experts, and business leaders exchange knowledge on Knowledge Graphs and Artificial Intelligence. It is a premier space for exploring applications in Neurosymbolic AI, Machine Learning, NLP, and Enterprise Data Management, turning cutting-edge research into practical, real-world solutions.

SEMANTiCS 2026 Blue Sky Winners

First Place

From Fact to Feeling: Narrative Knowledge Graphs for Human-AI Interaction

Anna Sofia Lippolis, Luiz Do Valle Miranda

Second Place

Can a Knowledge Graph Fall Down? Embodied Knowledge Graphs for Humanoid Robots

Blerina Spahiu, Loris Roveda

Third Place

From Sequential Language to Structural Intelligence: Toward Relationship-Aware AI

Alex Romanova

Want to read these papers? They are open access, hosted alongside the SEMANTiCS proceedings. Check them out here!

Please join us in congratulating the SEMANTiCS 2026 winners! 

CCC sponsors many Blue Sky tracks at various conferences in computing every year. If you are interested in requesting sponsorship of a Blue Sky track at your conference, please email blueskyideas@cra.org. 

Requests need only include a brief description of the conference and a proposed list of program committee members for the track. For more information — including guidelines for conference program committees, recommendations for selecting winners, and logistics for issuing CCC-sponsored travel awards to the winners, as well as a sample call for papers for a Blue Sky Ideas track — visit our website.

Tune in to the CCC LinkedIn Showcase Page for updates and more Blue Sky Track announcements. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.

Beyond Code: Engineering Trustworthy Software Systems with AI at Scale was authored by Randal Burns (Johns Hopkins University), Sebastian Elbaum (University of Virginia), Gabrielle Allen (University of Wyoming), Nils Aschenbruck (Osnabrück University), Terry Benzel (University of Southern California), William Gropp (University of Illinois Urbana-Champaign), Rick Kazman (University of Hawaii), and Manish Parashar (University of Utah).

Read the Report Here

Critical Paths Forward for Researchers

The report identifies seven technical directions for the computing research community:

• Bridge the intent gap by developing models and workflows capable of interactive requirement elicitation, formalized specifications that capture quality attributes, and traceability mechanisms that record why design choices were made
• Pursue neuro-symbolic integration, pairing generative models with symbolic reasoning so that code and its formal proofs are produced together and verified automatically
• Develop cross-layer orchestration through higher-level languages designed for generative agents, formal interfaces for inter-agent communication, and IDEs that function as orchestration dashboards
• Make security intrinsic to generation through goal-driven security interfaces, secure-by-default models, and cross-AI validation that uses one model to probe another for vulnerabilities
• Establish training data transparency and IP governance, including membership inference, provenance tracking, and models architected for verifiable attribution
• Advance transpilation as a robustness strategy, forcing generated output into safe and verifiable representations and providing equivalence guarantees across hardware platforms
• Extend AI to deployment, developing agents that reason about package contents and dependencies, monitor federated environments, and remediate anomalies without manual intervention

Shifts Specific to Academia

Alongside these technical directions, the report recommends changes to how academic research in this area positions itself:

• Move industry partnerships past benchmark creation
  • Current industrial-academic alliances largely position academia as a supplier of evaluation benchmarks, limiting research impact and constraining the talent pipeline. A deeper partnership would give researchers the production context, telemetry, and scale that building and testing next-generation models and tooling now requires, and give industry a partner working on problems that sit behind corporate walls.
• Aim at the next generation of models rather than evaluating the current one
  • The report suggests building models from the ground up for formal verification, security-by-construction, and long-term maintainability, which is work that corporate labs routinely bypass under market pressure. University labs retain a lead in safety-related work, including the sociotechnical and ethical effects of AI, which gives them an advantage.
• Rebuild curricula around requirements, specifications, and system design
  • Because students will no longer build mental models through the act of programming, new pedagogical approaches are needed that cultivate system-level intuition and complexity management. Computational thinking itself also needs redefinition to include managing higher-level abstractions and anticipating the behavior of agentic systems.

Read the Full Report

The findings of this report are informed by discussions at the CCC Beyond Code: Engineering Trustworthy Software Systems with AI at Scale visioning workshop held February 25-26, 2026 in San Francisco, CA, generously supported by the National Science Foundation (NSF) and IEEE Computer Society. The workshop convened 41 experts across academia, industry, and government in artificial intelligence, software engineering, programming languages, cybersecurity, and systems engineering. 

We encourage all members of the computing research community to read the full report at the link below.

Read Beyond Code: Engineering Trustworthy Software Systems with AI at Scale

Tune in to the CCC LinkedIn Showcase Page for updates and more reports like this. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.

This material is based upon work supported by the U.S. National Science Foundation (NSF) under Award Nos. 2300842 and 2619366. These awards support the Computing Community Consortium (CCC), a programmatic committee of the Computing Research Association (CRA). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

The Computing Community Consortium (CCC) is pleased to announce the third webinar in the Community Chat series: Powering the Solution: Balancing Computing’s Energy Demands with Its Contributions to Sustainability. Join the authors of the recent

Guest Speakers Share Their Expertise

In this one-hour Zoom webinar and Q&A moderated by article author Michela Taufer (University of Tennessee Knoxville), hear key findings from the article alongside three short talks centered around different aspects of the sustainability conversation.

• Chandra Krintz, Professor of Computer Science at the University of California, Santa Barbara, will open with a look at agriculture and ecology, and what balancing computing’s demands against broader societal needs looks like in applied environmental research.
• Ana Radovanovic, Senior Staff Research Scientist and Technical Lead for Energy Analytics and Carbon Aware Computing at Google, will follow with carbon-aware scheduling and how industry is thinking through the same balance. She previously led the team behind Google’s Carbon-Intelligent Computing System, which shifts flexible computing workloads to times and places where cleaner power is available.
• Adam Wierman, Professor of Computing and Mathematical Sciences at the California Institute of Technology, will close with a talk on the community impact of data centers, and whether they are a burden or an asset to the communities that host them.

All speakers will then be available for questions from the audience during a Q&A, joined by the other authors of the article:

• Michela Taufer, University of Tennessee Knoxville
• Elizabeth Bradley, University of Colorado Boulder
• Tevfik Kosar, University at Buffalo
• Siddharth Parida, Embry-Riddle Aeronautical University
• Masood Parvania, University of Utah
• Eve Schooler, University of Oxford
• Bronis R. de Supinski, Lawrence Livermore National Laboratory

Join Us

Date: Thursday, October 8, 2026

Time: 3:00 PM ET

The question around computing’s energy usage has become a global one. This is your chance to hear directly from the experts and make your voice heard. Use the link below to register for the webinar, and feel free to submit questions ahead of time. 

Can’t make it? Register anyway and CCC will send you the recording afterward. We hope to see you there!

Register Here

Tune in to the CCC LinkedIn Showcase Page for updates and more Community Chats like this. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.

This material is based upon work supported by the U.S. National Science Foundation (NSF) under Award Nos. 2300842 and 2619366. These awards support the Computing Community Consortium (CCC), a programmatic committee of the Computing Research Association (CRA). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

Action is urgent to realize the full potential of non-military drones. The report establishes milestones for 2028, laying the groundwork for its 2035 vision now.

See Key Recommendations for the Future of Drone Computing

Drones as a National Infrastructure

Computing on the Fly envisions a decade in which drones move goods, medical supplies, and information at a scale comparable to national infrastructure investments like highways and the electric grid. Natural disaster detection drones could identify wildfire sources within minutes, medical supply chains could bypass ground congestion to reach rural hospitals, and nationwide fleets could continuously inspect bridges and power lines.
Standing between today’s capabilities and that future is what the report authors call a “capability gap,” where hardware and aspirations are outpacing the software and systems needed to operate safely at scale. Drone research also remains fragmented across communities, which has left the field without a unified vision and put national leadership in the drone space at risk. Closing the gap requires coordination as much as it needs technological advances, as no single sector can close it alone.

Recommendations for Four Communities

The brief organizes its guidance around four groups, each with a distinct role in closing the capability gap:

• Scientific funding agencies, on the shared research infrastructure and program structures the field needs
• Policymakers, on standards, certification, and regulation as technical problems rather than administrative ones
• Industry, on interoperability, security, and what it takes to operate reliably at fleet scale
• Academia, on workforce development and the cross-disciplinary work the field currently lacks

The report identifies significantly scaling domestic drone production as a central, and time-sensitive, goal for this vision. Many of the brief’s cross-cutting recommendations thus revolve around this aim and the different roles each group plays in allowing that to happen.

Read the Key Recommendations Brief

The findings in the full report, as well as this new brief, come from CCC’s 2025 Computing on the Fly: Navigating a Vision for the Future of Drone Computing workshop. The workshop — co-sponsored by the IEEE Computer Society — took place in Washington, DC in December 2025, convening experts across academia, industry, and government to envision the future of drone computing. These conversations were synthesized in the full Computing on the Fly report by lead authors Kevin Butler (University of Florida), Christopher Stewart (The Ohio State University), Nils Aschenbruck (Osnabrück University), Weisong Shi (University of Delaware), Ufuk Topcu (University of Texas), and Deborah Silver (Rutgers University).

To read the most critical recommendations from the report, read the brief below.

Read the Brief

Tune in to the CCC LinkedIn Showcase Page for updates and more resources like this. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.

This material is based upon work supported by the U.S. National Science Foundation (NSF) under Award Nos. 2300842 and 2619366. These awards support the Computing Community Consortium (CCC), a programmatic committee of the Computing Research Association (CRA). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

The role of the software engineer is fundamentally changing. AI-assisted software development has, in many cases, become a new norm, fostering a demand for unprecedented speed in production. Yet the transition from traditional coding to AI-powered development is not as simple as receiving a boost to productivity. It requires both brand-new skills on the part of the developer and a more fundamental rethinking of how software — especially at scale — is generated. In this rapidly evolving field, new understandings of how to capture user intent, ensure system quality at AI-generative scales, and train the next generation of developers are essential.

Beyond Code Report

Beyond Code: The Workshop

To better understand and address these challenges, the Computing Community Consortium (CCC) held its Beyond Code: Engineering Trustworthy Software Systems with AI at Scale workshop — co-sponsored by the IEEE Computer Society — from February 25-26, 2026 in San Francisco, CA. The workshop brought together experts in artificial intelligence, software engineering, programming languages, cybersecurity, and systems engineering to assess the evolving role of AI in developing complex and large-scale software systems. 

CCC’s newest report contains findings and recommendations from the Beyond Code workshop. The report is authored by Randal Burns (Johns Hopkins University), Sebastian Elbaum (University of Virginia), Gabrielle Allen (University of Wyoming), Nils Aschenbruck (Osnabrück University), Terry Benzel (University of Southern California), William Gropp (University of Illinois Urbana-Champaign), Rick Kazman (University of Hawaii), and Manish Parashar (University of Utah). It presents critical challenges and recommendations for the future of large-scale AI-generated software development.

Central Topics

Participants of the Beyond Code workshop worked to identify current gaps and struggles in the current capacity to generate large-scale AI-generated software systems. In addition to those findings, they collaborated to find the best ways forward given the rapid changes taking place in the area, setting out concrete recommendations for the future of the field. Some of the topics central to the report include:

• The intent gap: natural language as a lossy interface

Natural language is the primary interface to code models, but its ambiguity and lack of formal semantics lose information in translation. 

• Machine representations beyond human-centric syntax

First-generation models are constrained by syntax designed for human readability, which forces inefficient trial-and-error generation loops.

• Quality attributes and formal assurance

Models produce functional logic but fail on performance, security, and numerical accuracy. Generation velocity and volume outpace human auditing capacity and make traditional quality metrics obsolete. 

• Security as an intrinsic property

Goal-driven security interfaces and secure-by-construction generation, rather than security applied after the fact.

• Multi-agent orchestration and deployment

Operating teams of agents complicates things. It introduces coordination friction, composability failures across siloed outputs, and integration liabilities at deployment. 

• Design rationale, maintainability, and legacy systems

As developers distance themselves from code generation, the rationale behind design choices is lost. Codebases become confusing and the capacity to maintain legacy systems diminishes.

• The developer role, deskilling, and education

Generative tools amplify existing productivity differences. They can enhance expert capability while also creating intellectual dependency and deskilling among the less-experienced. The role of software engineer itself is changing to be a multi-agent orchestrator. Curricula and training must follow.

• The research ecosystem

With industry commanding frontier models and compute infrastructure, academia risks relegation to post-hoc evaluation of commercial artifacts.

• Training data transparency and IP governance

Frameworks for data transparency and IP governance for AI-generated code, including membership inference and provenance tracking.

The report poses recommendations for tackling these pressing challenges within the computing research community.

Read the Full Report

The Beyond Code: Engineering Trustworthy Software Systems with AI at Scale report is available now. Its recommendations aim to create an ecosystem where large-scale AI-generated software is complex, interdependent, and, above all, trustworthy. We encourage all members of the computing community to read the full report here.

Tune in to the CCC LinkedIn Showcase Page for updates and more reports like this. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.

This material is based upon work supported by the U.S. National Science Foundation (NSF) under Award Nos. 1734706 and 2300842. This award supports the Computing Community Consortium (CCC), a programmatic committee of the Computing Research Association (CRA). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

To that end, the Computing Community Consortium’s (CCC) newest report, Computing on the Fly: Navigating a Vision for the Future of Drone Computing, lays out recommendations for maximizing the potential of drone computing. The findings are informed by the December 2025 Computing on the Fly workshop, which convened experts across computing research, industry, and government to craft an informed and realistic, yet visionary, roadmap for ensuring future U.S. leadership in this competitive field.

Recommendations for Funders of Scientific Research

Near-peer competitors are out-investing the U.S. in autonomous systems’ integration, threatening American leadership in a foundational technology advancement. Cross-agency coordination is essential to maintain American leadership in this area. 

In practice, this coordination looks like the National Science Foundation (NSF) leading foundational research while mission agencies like the U.S. Department of Agriculture (USDA), Department of Transportation (DoT), Federal Aviation Administration (FAA), Federal Communications Commission (FCC), and Department of Defense (DoD) ensure applications address real-world constraints and national priorities. 

Equally critical is establishing a national testbed consortium enabling reproducible research across diverse platforms and environments, ideally serving 20 universities and 100+ industry partners. And further, it is necessary to specifically fund integration platforms that compose AI agents into safety-critical systems.

Recommendations for Policymakers

For non-military drones to realize their full potential, policy frameworks must be proactive rather than reactive to the rapidly evolving technology. It is essential to adopt these reactive frameworks across operator licensing and training, vehicle certification and maintenance, and traffic management, spectrum allocation, and airspace design. 

At the same time, regulatory sandboxes are important for establishing safe experimentation frameworks. It is only through such experimentation that drones can enable solutions addressing issues like the $150B annual weather disaster burden while maintaining public safety. These regulatory sandboxes can accelerate innovation while managing risk. 

Most fundamentally, however, regulatory environments must be conducive to drone production and operation at scale. Initial stakeholder sessions detailing the current bottlenecks and barriers could inform critical decisions in the coming years.

Finally, international coordination is key to ensuring U.S. leadership in this strategic domain. Policy harmonization with allies prevents market fragmentation while maintaining security advantages over adversaries. 

Read the Full Report

The full report, including in-depth explanations and rationale for these recommendations, is available below. We highly recommend reading it, and subscribing below for CCC’s upcoming Key Recommendations Brief on this topic.

Read the Report Here

Tune in to the CCC LinkedIn Showcase Page for updates and more reports like this. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.

This material is based upon work supported by the U.S. National Science Foundation (NSF) under Award Nos. 1734706 and 2300842. This award supports the Computing Community Consortium (CCC), a programmatic committee of the Computing Research Association (CRA). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

The complete list of authors includes: 

• Michela Taufer, University of Tennessee Knoxville
• Chandra Krintz, UC Santa Barbara
• Elizabeth Bradley, University of Colorado Boulder
• Michael Crowley, University of Southern California
• Tevfik Kosar, University of Buffalo
• Siddharth Shiladitya Parida, Embry-Riddle Aeronautical University
• Masood Parvania, University of Utah
• Eve Schooler, University of Oxford
• Utkarsh Shah, Google
• Bronis R. De Supinski, Lawrence Livermore National Laboratory
• Adam Wierman, California Institute of Technology

See the registration at the bottom of this page to hear directly from these authors in an upcoming CCC Community Chat webinar and Q&A.

CCC Task Force Origins

CCC task forces are groups of computing researchers, composed of CCC Council Members and outside experts, who have identified a long-term computing challenge and aim to produce a resource to help guide the community. In 2023, when Michela Taufer and Adam Wierman joined the Computing Community Consortium (CCC) Council, the first discussions of a piece like “Balancing Sustainable Computing and Computing for Sustainability” were already taking place. At the time, Elizabeth Bradley and Chandra Krintz were part of a CCC task force on sustainability in computing; with the additions of Michela and Adam, they worked together to define the scope of the work. 

Partly, this meant reconciling the fact that the very same computing advances that often support sustainability work and research also impact the environment through their significant energy demands. They knew that they wanted that tension to be front and center in their work, and that they wanted the input and participation of members of the broader computing community on this topic. 

To that end, they ultimately invited other members of the community to join them in speaking at two different CCC-funded panels: one at the International Conference for High Performance Computing, Networking, Storage, and Analysis (SC24) in Atlanta, GA in 2024, and the second at the American Association for the Advancement of Science (AAAS) 2025 Annual Meeting in Boston, MA. They supplemented the discussion with panelists from different research backgrounds, as well as connected directly with two different audiences key to the conversation: the developers of high-energy computing technologies and advances, as well as their users.

In Michela Taufer’s words, “We had these visions and these ideas, but we had to find a way to come together and coordinate so that what we were doing became a roadmap for the whole community. And it was CCC that enabled us to come together to do that.”

Balancing Sustainability

In the end, the authors draw an important distinction between sustainable computing and computing for sustainability in the CACM article. Sustainable computing involves minimizing the environmental footprint of digital infrastructure — the digital infrastructure that, notoriously, has led to the proliferation and impact of widespread data centers. Computing for sustainability, on the other hand, means applying computational tools to solve pressing environmental sustainability challenges. The article argues that both approaches are critical for improving computing’s impact on the world, and that coordinated multidisciplinary action — involving computing research, technology development, and governance — is the key to striking that balance. 

For author Adam Wierman, that balance is not only achievable, but deeply urgent. “The energy demand of AI has already outrun what the grid was built to deliver,” he said. “Coordinating computing with the grid by shifting work in time and place is no longer optional.”

CCC’s mission of convening across all sectors of the computing research community also had a significant impact on the ultimate direction of the article. Chandra Krintz highlighted the necessity of working together across disciplines and sectors to make their vision a reality, and how CCC uniquely positioned them to make those connections: “Computing has a unique dual role in sustainability: we must reduce the energy and resource demands of the technologies we create, while harnessing those same technologies to address some of society’s most pressing challenges. Neither goal can be achieved by one discipline alone. Through CCC, we have been able to bring together perspectives from across computing, energy, science, policy, and industry to begin building a shared path forward.”

Join the Conversation

Want to learn more about balancing sustainability and hear directly from the article authors? Join the upcoming CCC Community Chat! 

Powering the Solution: Balancing Computing’s Energy Demands with its Contributions to Sustainability is a one-hour webinar and Q&A that will take place on Thursday, October 8, 2026 at 3 p.m. ET. This is your chance to better understand how computing can contribute to a more sustainable world and ask the authors, who are experts in the field, any burning questions you may have. Register below!

Register Here

Tune in to the CCC LinkedIn Showcase Page for updates and more reports like this. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.

This material is based upon work supported by the U.S. National Science Foundation (NSF) under Award Nos. 1734706 and 2300842. This award supports the Computing Community Consortium (CCC), a programmatic committee of the Computing Research Association (CRA). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

The renewal means CCC’s visioning activities with the computing research community will continue, supporting the evolving field through 2030 on this new grant.

CCC’s Mission

What are the next big computing research challenges — the ones that will define the future of the field, galvanize researchers, and spark investment and public support?

CCC mobilizes the computing research community to address these challenges. Since 2006, CCC has served as a bridge between computing experts across disciplines and sectors, convening them for visioning activities where they are asked to not only envision the future of computing, but also the path there. These collaborative efforts between academia, industry, and government — and the reports, white papers, and other resources that result from them — leverage the collective expertise of the field to set long-term computing research trajectories in motion and address emerging national and global challenges. 

What to Expect Over the Next Four Years

• Visioning workshops. These invitation-based, multi-day workshops assemble a group of experts around a specific research frontier. Topics originate from the CCC Council or from the community through CCC’s visioning proposal process. We encourage any member of the computing research community to submit a proposal!
• Reports, briefs, and white papers. CCC reports and briefs bring the findings of those workshops to the community. The recommendations within them inform researchers, industry folks, and policymakers about the focused work and funding required to serve the future goals of the field. CCC white papers do similar work, providing community-based expertise and recommendations without pulling from a fully-fledged workshop effort.
• Responses to federal Requests for Information (RFIs). When federal agencies release RFIs on computing research priorities, the CCC Council, working with CRA Government Affairs, submits a consolidated response. This is one of the most direct paths for community expertise to inform federal policy.
• Community Chats. CCC’s webinar series brings report authors, workshop organizers, and subject matter experts into conversation with a broad audience. These Community Chats span wide ranges of subjects within computing, are open to everyone, and are recorded for later viewing.
• Medium posts. CCC’s presence on Medium brings big computing questions to light. Expect accessible writing on computing findings and emerging topics.
• Blue Sky track sponsorships. CCC sponsors Blue Sky Ideas tracks at conferences across computing research, creating space for visionary and forward-looking papers. Sponsorship includes travel support for winning authors.

Get Involved

Two CCC projects are already on the books under this new award. These include our upcoming Full Stack Sustainability: From Silicon to Software workshop in Austin, TX — co-sponsored by the IEEE Computer Society — as well as an open Community Chat on October 8, 2026 called Powering the Solution: Balancing Computing’s Energy Demands with its Contributions to Sustainability (register here). Additional future visioning topics in the works include agentic AI for hardware design and quantum computing.

CCC strongly encourages members of the computing community to subscribe here and follow the CCC LinkedIn Showcase page to stay in the loop about our latest resources, events, and ways to get involved. For individual expert perspectives and commentary on computing that may not fit in a workshop report, follow CCC on Medium.

CCC looks forward to another four years of visioning with the inspiring computing research community!